Andrew, your logic in "Slippery savings" seems a little slippery itself. If I'm understanding you correctly, you are missing something when it comes to the real TCO if there is a hardware failure.
You state that there are 10 virtual servers per VMware instance. But you say that should there be a hardware failure, a cost is only attached to one of the servers. Therefore, either the rest of the servers have no value or additional costs should be allocated to hardware? If there is such a catastrophic failure, 10 servers will be affected, not just one.
I do not believe there is a flaw in my argument, I think that perhaps I just did not spell things out clearly enough. Certainly if one host server were to go offline that would be extremely bad, resulting in the loss of the VMs on it. However, read this carefully:
"Even if a hardware failure does occur, the right planning will result in zero downtime, saving l o s t c r e a t i o n s lost sales."
I am arguing that if you employ such technologies as attached storage for VMware then you do not need to suffer downtime at all because you can just load the VMs on another host server. This is not possible with physical hardware which is why it incurred downtime recovery costs (there are some exceptions, and I can address them if you wish).
Even if you still believe that there is a flaw in my argument, I do not think it is fatal. Instead of $0 downtime recovery costs, just figure in $2500 + (2 x $125) for the host servers and you get $2750. That makes the annual cost of hosting virtual services only slightly more expensive at $44,590. This is still compared to the annual cost of hosting 20 physical server of $110,100. Virtual services still comes out on top.
If I have misunderstood the comment please let me know. I am always open to feedback. SearchServerVirtualization.com labels me an expert, but really I am just another tech-head who has a passion for virtualization.
Nothing makes me any better than anyone else out there who works with the technology, and I will always listen to what others have to say.
This was first published in June 2006